KAPy – a community-based approach to the production of climate services

Climate services provide tailored information to support climate adaptation at the local level. One common form of climate service is the provision of downscaled climate projections, often bias-corrected using local observations and customised to meet the needs of local society based on extensive stakeholder engagement. Well-established examples of such services already exist in many European countries, while others are currently in the process of developing their own. While each such instance has its own peculiarities, there is also a high degree of overlap and duplication between these services. For example, the Danish “Klimaatlas”, the Norwegian “Climate in Norway 2100” report and the Swedish “Advanced Climate Change Scenario Service” all start from the EURO-CORDEX ensemble, bias-correct against local datasets, and produce comparable indicators, albeit entirely independently of each other. Recognising the potential to reduce duplication, to learn from each other and to enable the development of climate services in new regions, we established the KAPy (Klimaatlases in Python) project and network. KAPy builds on an open-source software stack centred on the Python language, leveraging the extensive tools already developed in this programming community. The use of workflow control tools from the field of bioinformatics enables reproducibility and scalability, while the open-source approach drives both collaboration and transparency. We illustrate the capability of this tool to produce climate service information using, as an example, ongoing work in Ghana, with a detailed analysis of the efforts required to produce climate-service ready indicators starting from scratch: after downloading of data was complete, configuration, bias-correction and indicator production was possible in a matter of a few hours. We also detail experiments showing how bandwidth limitations can be circumvented using cloud computing, further increasing the productivity and enabling implementation in resource limited situations, such low and low-middle income countries. We conclude with an open invitation to all to join the KAPy network as both users and developers, and thereby contribute to making climate services more transparent and widely accessible.